Male and female WKY rats for both cohorts were purchased from Charles River Laboratories (Raleigh, NC). Animals were housed in the U.S. EPA Office of Research and Development Research Triangle Park animal facility accredited by the Association for Assessment and Accreditation of Laboratory Animal Care at atmospheric housing conditions of 21°C ± 1°C, 50%–65% relative humidity under a 12 h light/dark cycle. Rats were pair-housed in polycarbonate cages containing hardwood chip bedding and Enviro-dri wrinkle paper for enrichment, and provided with Purina (5,001) rat chow (Brentwood, MO) and water ad libitum, except during exposure. Exposure occurred at 12–13 weeks of age, and males weighed ∼250–300 g and females weighed ∼150–200 g. Sexes were housed and exposed separately. All experimental procedures received prior approval from the U.S. EPA Health Institutional Animal Care and Use Committee.

Data presented here for nasal and bronchoalveolar lavage injury markers were collected from our recent acrolein exposure study, cohort 1 (Alewel et al., 2023b). A new cohort of male rats (cohort 2), used for collecting tissues for transcriptomic analysis, underwent exact same exposure conditions. In brief, rats for each cohort were exposed to air or acrolein (n = 8/sex/group) (exposures for each sex conducted separately). Over 2 consecutive days prior to exposure, rats were acclimated to nose-only inhalation tubes (Lab Products, Seaford, DE). On day 1, rats underwent two 2 h acclimations 4 h apart, and on day 2, rats underwent a single 4 h acclimation. On the day of exposure, veterinary-grade artificial tears (Akorn Animal Health, Inc., Lake Forest, IL) were applied to the eyes and rats were exposed via nose-only inhalation to air (0 ppm) or acrolein for ∼4 h. Previously, to assess in real-time acrolein concentration-related effects on respiratory parameters during the 4 h exposure period, head-out plethysmography (HOP) was performed using a graded half-log exposure concentration paradigm, where for the first 30 min rats were exposed to 0, 0.10, 0.316, and 1 ppm acrolein (7.5 min/concentration) followed by 3.5 h of exposure at 3.16 ppm in the same animals. A separate air control group was also included that was exposed in parallel to filtered air for 4 h. Plethysmography, serum/plasma, and some nasal lavage fluid (NALF) and bronchoalveolar lavage fluid (BALF) data are reported in detail in our prior publication (Alewel et al., 2023b). To be consistent with that exposure paradigm, in this study for collecting nasal and lung tissue, the same protocol was used. Necropsies for sample collection in each cohort occurred immediately after exposure. Although the highest concentration of acrolein used here (3.16 ppm) is higher than ambient traffic-related acrolein pollution levels (De Woskin et al., 2003), woodsmoke and mainstream cigarette smoke contain up to 50 ppm and 90 ppm acrolein, respectively (Einhorn, 1975; Hristova et al., 2012).

Acrolein was obtained as a gas stored under inert nitrogen for transport (Airgas, Morrisville, NC). For exposures, mass flow controllers modulated flow of acrolein and dilution air (purified air) to generate desired concentrations using a modified in-house gas blending system (MKS, Andover, MA). Acrolein was directed to flow through a 52-outlet port nose-only inhalation tower with air flow of 0.35 L/min per inhalation port. Targeted acrolein concentrations delivered to nose-only ports in exposure tower were continuously measured during exposure using an Agilent 6890 gas chromatography analyzer with flame ionization detection and 624 capillary columns (Supelco, Bellefonte, PA), and flows were adjusted as necessary to maintain the targeted acrolein concentrations (no deviations in target concentration were noted). Temperature (21.8°C ± 0.3°C) and relative humidity (38.9% ± 2.7%) in the room where the nose-only chambers were located were monitored once per hour during exposure.

Immediately after exposure, rats were transported for necropsy to be weighed and euthanized via an intraperitoneal injection of sodium pentobarbital (>200 mg/kg) (Fatal-plus; Covetrus, Portland, ME) in a staggered manner between air- and acrolein-exposed rats (all necropsies completed within 1-2 h after end of exposure). For the animals in the first cohort, BALF was collected. The left lung lobe was tied off, and calcium- and magnesium-free phosphate buffered saline (PBS) warmed to 37°C was instilled into the lung through the trachea. BALF instillation volume was based upon a total lung capacity being 28 mL/kg body weight and right lung weight as approximately 60% of total lung weight. NALF was collected from the same cohort of animals through a retrograde instillation of 2 mL warm PBS through the pharyngeal opening of the nasal cavity, allowing lavage fluid to freely drip through nares. Another 1 mL of air was instilled in the same manner to collect any residual NALF. Aliquots of whole NALF and BALF were used for total cell quantification, using a Z1 Coulter Counter (Coulter, Inc., Miami, FL). Albumin levels in NALF and BALF were assessed using kits from Sekisui Diagnostics (Burlington, MA), with the protocol for this assay adapted for use on a Konelab Arena 30 clinical analyzer (Thermo Chemical Lab Systems, Espoo, Finland). From the second cohort of animals, un-lavaged left lung was removed and snap frozen in liquid nitrogen. From the same cohort, the head was removed and split down the sagittal crest to reveal the nasal cavities. Lateral and septal wall nasal respiratory epithelium was scraped from the nasal cavity, collected on parafilm, and frozen in liquid nitrogen. Olfactory epithelial tissues, which are visually distinct, were avoided while collecting septal and lateral wall respiratory epithelial tissues. In rodents, the olfactory mucosa is anatomically blocked from main inspired airflow (Harkema, 1990; Chamanza and Wright, 2015). Therefore, to not introduce variability in mRNA assessment that may arise from differential expression profiles between structurally and functionally unique nasal mucosal zones, we focused our assessment on nasal respiratory epithelium. All samples were stored at −80°C for later RNA analysis.

NALF and BALF cytokines interleukin-6 (IL-6), tumor necrosis factor-α (TNF-α), C-X-C motif chemokine ligand 1 (CXCL1), interleukin-1β (IL-1β), interferon-γ (IFN-γ), and interleukin-13 (IL-13) were analyzed using V-PLEX pro-inflammatory panel 2 rat kits (Mesoscale Discovery Inc., Rockville, MD). The manufacturer’s protocol was followed for assessment in multiplex format. Electrochemiluminescence signals were measured using MESO QuickPlex SQ 120 platform (Mesoscale Discovery Inc., Rockville, MD). Protein signals were measured using a 4-parameter logistic model to determine sample concentration using MSD Discovery Workbench analysis software (Mesoscale Discovery Inc., Rockville, MD).

Nasal epithelial tissue (n = 5-6/group) and uniform portions of the left lung (n = 6–8/group) from male rats were used for RNA extraction and processed with RNeasy mini kits (Qiagen, Valencia, CA) as directed by the manufacturer. Following extraction, RNA quantity and purity (260/230 and 260/280 ratios) were determined using a Nanodrop 1000 (ThermoFisher Scientific Inc., Waltham, MA). RNA integrity was assessed by the RNA 6000 LabChip^® kit and a 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA) using a RIN cutoff of 7. mRNA samples were randomized and processed on an Apollo324 automated system (Takara Bio Inc., Kusatsu, Japan) for library prep with PrepX mRNA 48 protocol v19, using the PrepX^TM RNA-seq for Illumina Library Kit (Takara), SuperScript III reverse transcriptase (ThermoFisher), and AMPure XP Beads (Agilent). PCR amplification with 48 index primers was run for 16 cycles and resulting PCR product quality was analyzed again via Qubit (ThermoFisher) and Bioanalyzer (Agilent). An RNAseq library was prepared with Wafergen’s PrepX mRNA 48 protocol and dsDNA products were prepared from cDNA. Each library was sequenced according to Illumina NextSeq 500, with a final concentration of 2.2 pM + 2% PhiX and run for 75 cycles (Illumina Inc., San Diego, CA).

The data for NALF and BALF albumin, total cells, and cytokine proteins were analyzed using GraphPad Prism software v9. Due to low basal levels of proteins in NALF, some samples for albumin content in air-exposed groups were below the detection limit (BDL). For these samples, BDL values were imputed with the limit of detection (LOD) using LOD/√2. Likewise, in some cytokines, particularly among air-exposed controls in nasal lavage, few samples that were BDL were imputed in the same manner using LOD/√2 for each analyte for statistical comparison. First, outliers were identified and discarded using a robust regression outlier test (ROUT) with a false discovery threshold of 0.01 (Q = 1%). To meet the assumptions of normality and homoscedasticity within an analysis of variance (ANOVA), data underwent Shapiro-Wilks normality (α = 0.05) and Levene’s equality of variance tests. Data that did not meet ANOVA assumptions of equal variance and/or normality were Log transformed [Y = Log (Y)] prior to analysis. NALF and BALF markers were analyzed separately using a two-way ANOVA (sex, exposure). A Tukey’s multiple comparisons post-hoc was used and significance was determined at p ≤ .05.

Sequenced mRNA reads were mapped to the rat genome (rn7) using ensemble release 105 in the Partek Flow suite, R version 4.1.3. In total, 24,091 genes were mapped. Genes that did not have at least 1 read in 10 samples were removed, leaving 20,580 genes. An average of 15.5 million reads were mapped per sample with a standard deviation of 5.5 million. One nasal sample was removed because of low total reads (∼300,000). Principal components analysis was performed to identify any samples that did not cluster and should be removed; none were removed in this step. The remaining samples had 15.8 million reads per sample with a standard deviation of 5.3 million, with an average read per gene of 766. Differential gene expression was assessed in DESeq2 package in R (v1.34.0) (Love et al., 2014) and normalized reads were assessed for acrolein effects. Differential expression analyses were performed within tissue to assess the effects of acrolein. Gene expression was considered different when the fold change (FC) was greater than 1.2 (upregulation) or less than 0.833 (downregulation) and the Benjamini–Hochberg false discovery rate ≤ 0.1. Sequencing data is deposited in the National Center for Biotechnology Information Gene Expression Omnibus under accession number GSE247698.

For each heatmap, normalized read counts for each gene were averaged and calculated as the standard deviation from the mean, reflecting row z-score of each gene as up- (yellow) or downregulated (blue). Genes were clustered according to average-linkage Euclidean distancing using the online tool http://heatmapper.ca (last accessed 22 June 2023). We utilized Ingenuity Pathway Analysis (IPA) software (QIAGEN, Redwood City, CA; www.qiagen.com/ingenuity; last accessed 22 June 2023), which compares our mRNA expression dataset to previously reported databases and reports overlaps in expression changes for interpretation of major pathway and gene changes. Top canonical pathways were selected based upon significance between overlap of molecules in our dataset and molecular changes consistent with a particular network (adj. p-value <0.1). To infer involvement of upstream regulators, we determined an |activation z-score| ≥ 2 for meaningful prediction of network connectivity with over- or under-expressed transcripts (Krämer et al., 2014).

To assess overall upper and lower airway injury and inflammation patterns in response to acrolein inhalation, we measured albumin and total cells in NALF and BALF (Figure 1). A single exposure to acrolein resulted in higher NALF albumin levels in male and female rats when compared to respective air groups, but did not change BALF albumin levels (Figure 1A). Interestingly, although both sexes demonstrated nasal protein leakage, only acrolein-exposed males had higher NALF total cell counts, which we previously reported as neutrophilic and eosinophilic inflammation (Alewel et al., 2023b) (Figure 1B). Further, acrolein exposure significantly elevated BALF total cell counts in both sexes (Figure 1B).

NALF and BALF samples were further assessed to examine the potential involvement of various pro-inflammatory cytokines (Figure 2). The only cytokine impacted in both sexes was IL-6, which, compared to air group, was higher in NALF of acrolein-exposed males and females and BALF of acrolein-exposed males, and nearly significant in female BALF (p = 0.053) (Figure 2A). Other cytokine alterations followed a sex-specific response only in NALF, suggesting an effect at the site of nasal acrolein deposition. TNF-α, IL-1β, and IFN-γ levels were higher in NALF of acrolein-exposed males, whereas no other changes in these cytokines were noted in NALF of females or BALF of either sex (Figures 2B, D, E). CXCL1 and IL-13 NALF and BALF levels were not impacted by acrolein exposure in either sex, except for a decrease in BALF CXCL1 in acrolein-exposed females (Figures 2C, F). Few cytokines showed differences among air control groups, where NALF IFN-γ levels and BALF CXCL1 levels were higher in female air controls compared to males (Figures 2C, E).

To assess nasal versus lung transcriptional responses to acrolein, we performed global mRNA analysis in each tissue. Only males were used due to their greater susceptibility to acrolein-induced airway injury and inflammation when compared to females. Figure 3 shows volcano plots of RNAseq data obtained from nasal and lung tissues. Acrolein inhalation changed expression of 452 genes in the nasal epithelium (310 upregulated and 142 downregulated) (Figure 3A). Comparatively, acrolein exposure resulted in nearly 5-fold fewer DEGs in lung, changing expression of 95 genes (80 upregulated and 15 downregulated) (Figure 3B). To determine the overall effects on biological processes within nasal and lung tissue, IPA core analysis was performed. Figures 3C, D show the graphical summary of the interactive influence on key signaling regulators up- or downregulated after acrolein exposure in the nasal and lung tissue, respectively. This core analysis illustrates relations among predicted biological pathways, transcription factors, and important signaling modulators being impacted by acrolein exposure in the nasal versus lung tissue. Overall, tissue-specific differences can be seen in the number and function of biological processes involved in transcriptional alterations (Figure 3). This IPA core analysis demonstrates that cell signaling pathways and transcriptional activations involving innate inflammatory processes mediated by oxidative stress are induced in the nasal tissues with inhibition of IL-10 signaling. The increases in mRNA transcriptions for IL-6, IFN-γ, IL-1β, and TNF genes (Figure 3C) coincide with increases in these cytokine proteins in the NALF. The absence of these pathways activated in the lung is consistent with the lack of increases in these cytokines in the BALF. Rather, the pathways changed in the lung tissue show that cell growth and proliferation processes might be activated (Figure 3D).

To highlight the similarities and differences of acrolein effects on cellular processes between the nasal and pulmonary tissues, the top 20 up- and 20 downregulated genes in nasal tissues were compared with same genes in the lung and vice versa (Figure 4). The heat map in Figure 4A displays the top 20 DEGs up- and downregulated in the nose (ranked based on fold change), with the listing of changes in same genes in the lung. In the nose, the most upregulated gene was Hmox1 (Log2FC = 4.63), a gene regulating iron homeostasis and oxidative imbalance (Chen et al., 2020), and the most downregulated gene was Gmnc (Log2FC = −1.40), a gene involved in the differentiation of ciliated epithelial cells (Zhou et al., 2015) (Figure 4A). Other top upregulated nasal genes included multiple FOS family genes (Fos, Fosl1, and Fosb) involved in acrolein toxicity (Yeager et al., 2016), and those downregulated included genes regulating ion channel functions and cellular transport (Figure 4A). It is noteworthy most of these same genes did not respond to acrolein exposure in lung tissue. While acrolein exposure caused only a fifth of the number of genes to be differentially expressed in the lung when compared to nasal airways, Hmox1 was commonly upregulated in both tissues (Figure 4B). The heat map in Figure 4B displays the top DEGs in the lung from acrolein exposure, with the same genes selected and matched for nasal airways for comparison. In the lung, the most upregulated DEG was Slc16a5 (Log2FC = 1.98) involved in transport of monocarboxylate metabolites such as pyruvate and lactate (Ren et al., 2022), and the most downregulated gene was Card9 (Log2FC = −0.95), an upstream regulator involved in activation of NFκB and apoptosis signaling (Bertin et al., 2000). This pattern of change depicts that in the lung, acrolein transcriptional response likely did not involve the activation of NFκB signaling despite observed increases in BALF cells. Other genes downregulated in the lung included Hpse2, Ahr, and Irs1, all involved in cellular metabolism. Overall, few overlapping genes were differentially expressed in both tissues following acrolein exposure. We found 7 overlapping genes, Hmox1, Cebpd, Akr1b10, Slc7a11, Slc16a5, Hap1, and Tsku, differentially expressed in both nasal and lung tissue (Supplementary Table S2).

To obtain insight into processes affected by acrolein exposure, we performed canonical pathway analysis of mRNA expression data. Table 1 shows the top 15 most significantly altered pathways (adj. p-value <0.1) in each tissue following acrolein exposure. Notable pathways enriched by acrolein exposure in nasal tissue included acute phase response (APR) signaling, Nrf2-mediated oxidative stress, IL-6 signaling, unfolded protein response, ERK5 signaling, and HMGB1 signaling. Pathways downregulated by acrolein exposure in nasal RNAseq data were IL-10 and ATM signaling (Table 1). Unlike widespread transcriptional changes indicative of oxidative stress response and activation of inflammatory processes in the nasal tissue, the enriched canonical pathways in the lung indicated changes in xenobiotic metabolism.

Figure 5A shows 18 significant DEGs within the APR network in nasal tissue. Multiple genes associated with APR activation were enriched in nasal but not lung tissue, including Il6, Cxcl2, Hamp, Tnfrsf1a, and Serpine1, corroborating NALF and BALF pro-inflammatory cytokine involvement (Figure 5A). Genes Map2k3 and Map3k14, components of mitogen-activated protein kinase (MAPK) cascades in innate immune activation, were significantly enriched in nasal epithelium (Figure 5A). Acrolein’s effect on APR genes in the lung was smaller than that in the nasal tissue. Further, 18 DEGs comprised Nrf2-mediated oxidative signaling in the nose (Figure 5B). Various AP-1 subunit genes (Fos, Fosl1, Jun, Junb, and Jund) were enriched in the nose, along with Gclc and Gclm (Figure 5B). Among DEGs identified within these nasal pathways, only Hmox1 was significantly altered in the lung (Figures 5A, B).

To explore the involvement of nuclear receptor signaling in upper and lower airway acrolein effects, we identified molecules pertaining to glucocorticoid receptor (GR) and G protein-coupled receptor (GPCR) signaling through IPA analysis (Figure 6), which is potentially affected based on our prior observation that circulating corticosterone and epinephrine are increased in rats after acrolein exposure (Snow et al., 2017; Alewel et al., 2023b). Overall, acrolein-exposed rats exhibited many gene expression changes modulated by GR and GPCR signaling in the nose that were not observed in the lung. Prominent genes enriched in the nasal GR signaling pathway included Hsp70 encoding genes (Hspa1a/Hspa1b, Hspa1l, and Hspa8) and Sgk1 (Figure 6A), which is a glucocorticoid-regulated kinase prompted by cellular stress (Jang et al., 2022). Multiple GPCR signaling network changes were observed in the nose following acrolein exposure, whereas these effects were not seen in the lung (Figure 6B). GPCR-mediated changes altering intracellular adenylate cyclase or cyclic-AMP (cAMP) levels were noted through increased nasal expression of genes such as Adra2a, Adora2a, Creb3l3, and Atf4. Among the 21 GR and 26 GPCR signaling mRNA expression changes noted in the nasal tissue, none of these genes were significantly changed in the lung (Figure 6).

Because one of the objectives of our transcriptomic analysis was to identify the involvement of adrenal-hormone signaling in acrolein airways effects, we evaluated specific glucocorticoid-responsive genes that are extensively validated as consistent markers of GR activity due to their gene loci often falling within glucocorticoid response elements (GRE) (Wang et al., 2004; Kan and Himes, 2021). Table 2 shows 9 such genes selected from nasal and lung RNAseq data. Ccl2, Cebpd, Per1, Sgk1, Thbd, Ccl20, and Gadd45b were significantly enriched in nasal tissue only, except for Cebpd which shared increased lung expression. Interestingly, Fkbp5, a prominent modulator of glucocorticoid signaling, was enriched only in the lung, also seen in prior ozone studies (Henriquez et al., 2021).

We used the IPA upstream predictor tool to evaluate cytokines, transcriptional activators, or chemicals which shared similar gene changes with acrolein-induced airway effects (Table 3). Here, our gene expression dataset was compared to IPA databases to provide the most likely regulators responsible for network changes in either nasal or lung datasets. Nasal acrolein-induced changes predicted activated regulation by lipopolysaccharide (LPS), multiple cytokines, CREB1, and inhibition by various chemical and biologic agents (Table 3). Further, ozone and cigarette smoke were predicted chemical toxicants with gene signatures similar to acrolein-induced nasal effects, which is consistent with acrolein as a major tobacco smoke constituent (Hikisz and Jacenik, 2023), (Supplementary Table S6). Forskolin and methylprednisolone were also suggested upstream predictors of nasal transcriptomic changes (Supplementary Table S6), implicating the involvement of epinephrine-mediated GPCR signaling via activation of cAMP and corticosteroid effects on GR in acrolein-induced upper airway effects. The fewer transcriptional changes in the lung, however, were predicted to be related to different substances. The lung prediction algorithm identified similarities to activation of aflatoxin B1, NFE2L2 (Nrf2), and Vegf, as well as inhibition by multiple transcriptional regulators and chemical agents (Table 3).